ContentslistsavailableatScienceDirect
Sensors
and
Actuators
A:
Physical
jo u r n al hom e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / s n a
Wireless
deep-subwavelength
metamaterial
enabling
sub-mm
resolution
magnetic
resonance
imaging
Sayim
Gokyar
a,b,∗,
Akbar
Alipour
a,b,
Emre
Unal
a,b,
Ergin
Atalar
a,
Hilmi
Volkan
Demir
a,b,caDepartmentofElectricalandElectronicsEngineering,DepartmentofPhysics,NationalMagneticResonanceResearchCenter(UMRAM),BilkentUniversity, Ankara,TR-06800,Turkey
bUNAM-NationalNanotechnologyResearchCenterandInstituteofMaterialsScienceandNanotechnology,BilkentUniversity,Ankara,TR-06800,Turkey cLuminous!CenterofExcellenceforSemiconductorLightingandDisplays,SchoolofElectricalandElectronicEngineering,SchoolofPhysicaland MathematicalSciences,PhysicsandAppliedPhysicsDivision,NanyangTechnologicalUniversity,Singapore
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received9May2017
Receivedinrevisedform4March2018 Accepted19March2018
Availableonline20March2018 Keywords:
Magneticresonanceimaging Metamaterials
Deep-subwavelengthresonators
a
b
s
t
r
a
c
t
Awirelessdeep-subwavelengthmetamaterialarchitectureisproposed,modeledanddemonstratedfora high-resolutionmagneticresonanceimaging(HR-MRI)applicationthatisminiaturizedtoberesonantat approximately0/1500dimensions.Theproposedstructurehastheadjustableresonancefrequencyfrom
65MHzto5.5GHzforthesub-cmfootprintarea(8mm×8mmforthisstudy)andprovidesaqualityfactor (Q-factor)ofapproximately80infreespacefor123MHzofoperation.Thisstructureconsistsofacross-via metallizedpartial-double-layermetamaterial,sandwichingadielectricthinfilm;thisstructurestrongly localizestheelectricfieldinthisfilmandhasahighlycapacitivemetaloverlaythatallowsforawiderange offrequencyadjustment.Althoughtheachievedresonancefrequenciesenablealargenumberof appli-cations,asaproof-of-conceptdemonstration,weexperimentallyshowedtheoperationofthiswireless metastructureinHR-MRItohighlightitsprecisefrequencyadjustmentandsignal-to-noise-ratio(SNR) improvementcapabilities.TheproposedmetamaterialwasfoundtomaintainshighQ-factorsdespite loadingwithabody-mimickinglossyphantom.Theexperimentalresultsindicatedthattheproposed metastructurecanbeusedasanSNR-enhancingdeviceoffering15-foldSNRenhancementsthatallows forimagingofobjectsassmallas200mindiameterinitsvicinity,atanunprecedentedlevelof resolu-tionatthegivenDCfieldusingstandardheadcoils.Asaresultofitsdeep-subwavelengthminiaturization accompaniedbyreasonableQ-factorwithoutstandingresonancefrequencyadjustmentcapability,this classofmetastructureisprovedtobeanexcellentcandidateforinvivomedicalapplications.
©2018ElsevierB.V.Allrightsreserved.
1. Introduction
FollowingtheseminalworkofVeselago[1],metamaterialshave beenintroducedindifferentapplicationsincludingmaterial char-acterization[2],sensing[3–8],compactingdevices[9]andimaging ofsubwavelengthfeatures[10],spanningfromopticalfrequencies [11]totheradiofrequency(RF)region[12].Theoperating band-widthof thosemetamaterial devices hasbeentypicallynarrow becauseofthehighqualityfactor(Q-factor)elementsusedinthe unitcelloftheirstructures.Becausemagneticresonanceimaging (MRI)isbasicallyanarrowbandwidthimagingtechnique, meta-materialscouldbequiteattractiveforuseinMRIapplications.It
∗ Correspondingauthorat:DepartmentofElectricalandElectronicsEngineering, DepartmentofPhysics,NationalMagneticResonanceResearchCenter(UMRAM), BilkentUniversity,Ankara,TR-06800,Turkey.
E-mailaddress:sayim@ee.bilkent.edu.tr(S.Gokyar).
waspreviouslyshownthatmetamaterialstructurescanbeusedfor RFfluxguiding[13]andsignal-to-noise-ratio(SNR)improvement purposes[14].Thesemetamaterials,inadditiontoconventional imaging hardwareofthescanner,properlymanipulatestheEM fieldsintheirvicinitytoincreaseimagingsignalemittedfromthe imagedobjects(e.g.,tissues).However,thesewirelessdesignsare largerinsizetobeusedformedicalapplications;hence,thedesign ofwirelesssub-cmmetamaterialstoaddressinvivoMRIofsub-mm featureshasnotbeendemonstratedtodate.
High-resolutionMRI(HR-MRI) suffersthefundamental prob-lemofreducedSNRbecauseofthedecreasedvolumeoftheimaged voxels.ToincreasetheSNRofanMRimage,severalmethodscan beapplied.TheseincludeusingahigherDCmagneticfield[15], using highdensitycoil arrays withparallel imaging techniques [16,17],increasingthenumberofimageacquisitionandusing high-sensitivitywiredcoils[18–27].Practically,theDCfieldstrengthis predeterminedandassumedtobeconstantaftertheinstallmentof themainmagnet.However,usinghighdensitycoilarrayswith par-https://doi.org/10.1016/j.sna.2018.03.024
allelimagingtechniquesisoneofthemajorbreakthroughs;their performancescanbeincreasedbyusingwirelesscoilsforinvivo applications.Increasingthenumberofacquiredimagesincreases theimaging duration and total RF powerexposure of patients, whichisnotsuitablefor clinicalimaging practices.Hence,for a predeterminedDCfieldandareceivercoilconfiguration,increasing SNRwithoutincreasingthetotalRFpowerexposureandimaging durationbecomespossiblewithlocalizedcoilsolutions. Unfortu-nately,thesecoils[18–28]requirecomplicated electronics(e.g., matchingcapacitors,solder,diodesand/orcryo-coolingetc.)and theyareconnectedtothebodyofthescannerviaRFcables.Hence, theirapplication toinvivo operationsis inherentlychallenging becauseof RF heating risks [29]. Using wireless resonators for invivoMRIapplicationswouldincreasethedetectionperformance ofanMRIsystem[13,14,30–32].Therehasbeensignificanteffortin theliteraturetodecreasetheresonancefrequency(f0)of metama-terials[33]suchas,increasingthesizeoftheelementsgiveninthe slab[34]and/orusinglumpedcapacitors[12]tocapacitivelyload theunitcells.Althoughmostofthesemethodsareacceptablefor certainapplications,suchasstrainsensing[35],theyarenot suit-ableforinvivoapplicationsbecauseoftheirsize[34]andlumped capacitorsusedtotunetheirresonancefrequencies[12].However, noneofthepreviouswirelessdesigns,ormetamaterials,achieved theelectricalsizeofsmallerthan/300andaQ-factorofmorethan 50simultaneouslywithoutusingcryo-coolingoralumpedelement thusfar.
In this work, to address the aforementioned problems of HR-MRI,wedemonstrateawirelessdeep-subwavelength metas-tructureenablingaQ-factorofapproximately80infreespace,in ahighlycompactfootprintarea.Heretheproposedwireless meta-materialstructure,whilebeingelectricallyverysmall,isalsoshown tobeanexcellentcandidateforinvivoMRIapplicationsincluding HR-MRIatsub-mmresolution.
2. Methods
Theproposedstructureisacross-viametallizeddouble-layer metamaterial,inwhichtheconsecutivemetallayersarestrongly coupled (i.e. both inductively and capacitively) to each other throughanoverlayregiontodecreasetheresonancefrequencytoa pre-definedfrequency.Fig.1.ashowsaschematicoftheproposed structure. Conductive cross-via metallization results in a thin-film loaded semi-turn over-laid double-layer resonator. Unlike theclassicalsplitringresonators(SRR)andmulti-layerSRRs,this metamaterialstructureexhibitshigherinductivecouplingviathis cross-viametallizationbetweenconsecutivelayers.
2.1. Equivalentcircuitmodelingandfull-wavenumericalanalysis EquivalentcircuitmodelsofSRRs[36]andspirals[37]havebeen previouslyreportedintheliterature,wherethesinglelayer res-onatorsaremodeledasa seriesRLCcircuitwithpropermutual couplingterms.ResonancefrequencyofaresonatorisgivenbyEq. (1)
f0= 1 2
LeffCeff(1) whereLeffistheeffectiveinductanceandCeffistheeffective capac-itanceoftheoverallstructure.Q-factorofaseriesresonatorisgiven byEq.(2) Q=2f0Leff Reff = 1 Reff
Leff Ceff = f0 f3dB (2) whereReffistheeffectiveresistanceandf3dBisthe full-width-half-maximum(FWHM)bandwidthoftheresonator.UnlikepreviouslyFig.1.Schematicrepresentationoftheproposedwirelessmetamaterialstructure (notdrawntoscale).(a)Ametal-insulator-metal(MIM)devicewithcross-via met-allizationtoincreaseLeffinagivenfootprintarea(withsemi-turnoverlay).(b)The operationofthestructurecanbemodeledbyusingacascadedequivalentcircuit approach.Eachunitcelliscomposedofdifferentialinductance(dL),capacitance (dC)andresistance(dR)withamutualcouplingofdMforpartialoverlayregions.
studied structures, this metamaterial architecture is composed oftwoinductivelayersconnectedinseries(viacross-via metal-lization)withstrongcapacitivecoupling,resultingindistributed thin-filmcapacitancebehaviorthatcannotbeanalyzedusing con-ventionalmethods[40,41].
Beforewemodeltheproposedarchitecture,wecalculatedthe inductanceofasingleturnrectangularresonator,L0,byusingEq. (3) L0=0.6350rn2Dav
ln 2.07 +0.18+0.132 (3) andtheparallelplatecapacitance,C0,byusingEq.(4)C0=ε0εrA
d (4)
where0isthepermeabilityofthefreespace,ristherelative permeabilityofthematerialusedforresonatorfabrication(which hastobeunityforMRIoperation),nisthenumberofturn(unityfor asinglelayerstructure),Davistheaveragediameter(Dav=Do+D2 i), isthefillratio(= Do−Di
Do+Di),Doistheouterdiameter(sidelength
forrectangularresonators),Diistheinnerdiameter(side length-2×w)fortherectangularresonators),wisthemetallizationwidth, 0isthepermittivityofthefreespace,ristherelativepermittivity ofthedielectricusedforelectricfieldlocalization,Aistheparallel platesurfaceareaanddisthedistancebetweentheconsecutive layers(e.g.,dielectricthicknessfortheproposedstructure).
Finally,theACresistanceofthesingleturnstructureis calcu-latedbyusingEq.(5)
R0= l
wı(1−e−tmetal/ı) (5)
wherelisthemeantpathlengthofthemetaltrace(proportionalto thenumberofturns,n),tmetalisthethicknessofthemetallization, istheconductivityofthemetalusedforfabricationand␦isthe
skin-depthofthemetalforthegivenfrequencythatisformulated asgiveninEq.(6). ı=
2 2f (6)Stackingadditionalturnsdoesnotchangethefillratio,.Thus, WhenwesubstituteEqs.(3)and(5)intoEq.(2),weobservethat theQ-factorislinearlyproportionaltothenumberofturns,n,fora pre-determinedresonancefrequency,f0.Thisisnotthecasefor spiralresonatorsbecauseofdecreasingmutualcouplingamong consecutiveturns[33].
To model its behavior, we discretized the proposed struc-tureton0 unit cells,as depictedin Fig.1b;theseunitcells are cascadedtoconstructtheoverallstructure.Eachunitcellis com-posedofdifferentialinductance(dL=L0/2n0),differentialresistance (dR=R0/2n0),differentialthin-filmcapacitance(dC=C0/n0)and dif-ferentialmutualcoupling(dM,dM=kdL),wherekisthecoupling coefficientandhasavalueofapproximately1forstrongpositive couplingviacross-viaconnection[38].Forelectricallysmall struc-tures(e.g.,electricalsize<0/300forthiscase),thecornersofthe proposedstructurecanalsobemodeledbyusingtheaboveunitcell configuration.Similarly,theviametallizationisalsomodeledasa seriesRLcircuit(withoutdCelement)betweenconsecutivelayers. ByusingKirchhoff’scurrentlaw,weconstructedtheadmittance matrixasV[Y]=I,whereVandIarethevoltageandcurrentvectorsof thenodesrespectivelyandYistheadmittancematrixoftheoverall structure(detailsarereportedinthesupplementaryfile,sectionS1). TheresultingimpedancegraphsarecalculatedbyusingaMATLAB® (TheMathWorks,Inc.01760USA)todeterminetheresonance fre-quency(i.e.,calculatedfromthepeaksoftheimpedancegraphs) andtheQ-factor(calculatedusingthe,full-width-half-maximum, FWHM,ofthegraphs)ofthemetamaterialforvariousn0.
Theeffectsofvariousgeometricparameters,includingtdieland overlayarea,wereanalyzedbyusingafull-wavenumericalsolver, CST-MicrowaveStudioTM(CST,64289Darmstadt,Germany).The simulationdomainwascomposedofacubicvacuumenvironment withasidelengthof16mmandtheboundaryconditionsareset asperfectlymatchedlayers(PMLs)inalldirections.Asquarecoil havingasidelengthof8mm,atracewidthof1mm,anda met-allizationthicknessof35mispositioned0.2mmawayfromthe proposedstructuretobeusedasthepick-upcoiltomeasureits inputimpedance.Themetallizationmaterialusedwasgoldforboth theantennaandtheresonator.Thedielectricmaterialofthe pro-posedstructurewaschosenaslossypolyimidefromthelibrary ofCSTMicrowaveStudio,forfull-wavesolutions.Thefrequency domainsolverwasusedtoacquirethescatteringparametersof thewirelessstructure.Thissimulationenvironmentisdepictedin Fig.2.Here,wefollowedthemethodsdescribedbyGinefrietal. [39]andmeasuredtheinputimpedanceusinganetworkanalyzer (AgilentE5061B)andapick-upcoilantenna(detailsarereportedin thesupplementaryfile,sectionS2).
2.2. Microfabricationofmetastructures
Wefabricatedtwo differentdevices(rigid andflexible ones) withdifferentmicrofabricationmethods.Therigiddevicesaimto achieve thesmallestelectricalfootprint area(without targeting apredefinedoperationalfrequency)fora wirelessmetamaterial structure,where andtheflexibledeviceismicrofabricatedonto apolyimidesubstrateandtunedtoapredefinedMRIfrequency. Therigidsamples,presentedinFig.3(a),aremicrofabricatedonto a<111>siliconsubstrate.Byusingahard-maskwith complemen-tarySRRpatterns,wethermallyevaporatedtwo10mthickAu SRRlayers(heregoldischosenbecauseitisbiocompatible)witha
Fig.2. Schematicrepresentationofthemeasurementsetup(notdrawntoscale). Theresonatorisstronglycoupledtothepick-upcoilantennatomeasureits char-acteristicproperties.
Fig.3.Opticalphotographsofthemicrofabricateddevices.(a)Rigidsampleshave thefootprintof6mm×6mm(foreachfeature).Thesesamplesaremanufactured inarrayformontoasiliconwaferwitha1-mthicksiliconnitridedielectric sandwichedbetweentwo10-mthickgoldlayers.(b)Theflexiblestructureis microfabricatedontoapolyimidethinfilmwithsub-cmdimensions(8mm×8mm). 1mthickplasma-enhanced-chemical-vapor-deposited(PECVD) siliconnitride(Si3N4)sandwichedbetweeneachofthegoldlayers. Thesecondsample,presentedinFig.3(b),ismicrofabricated ontoan8mm×8mmfootprintareabyusingaflexiblepolyimide film(Kapton®)withaninitialthicknessof7.5m.Theproposed thinfilmisthinnedto2.7mbyusingreactiveionetching(RIE), with the following recipe: SF6:O2 of 45:15 sccm at 150W RF power,with25mTorrpressuretoincreasetheeffectivecapacitance betweenconsecutivelayers.Byusingahard-maskwith comple-mentarySRRpatterns,wethermallyevaporatedtwo10-mthick AuSRRlayerswithasymmetriccombsizes,withonelayeroneach sideofthepolyimidefilm,andintroducedviametallizationthrough thesubstratecross-connectingapairoftheoppositeedgesofeach SRRontheotherside.Subsequently,thesampleswereannealed at 250◦C for 5min for increased electrical conductivity.Unlike conventionalopticallithographytechniquesusedinrigidsample fabrication[40],thismethodpreventspotentialchemicalhazards forbiocompatibilityanddrawsthesimplestmethodologyforhigh yieldfabrications.
2.3. MRIsetup
To demonstrate the MRI operation of the proposed flexible metamaterialstructure,wepreparedagel-phantomwith dimen-sions80×80×40mm3byusing1g/LNaCl,2.5g/LCuSO
4and14g/L agarose-geltomimicthetissueproperties[41]witha correspond-ingrelativepermittivityofapproximately60andaconductivityof 0.5S/m.Twosetsofevenlydistributed13fiberpillars,eachpillar withadiameterof200m,wereimmersedintothepre-mentioned gelphantomandpositionedalong ˆz-directionwithdepthsof0.1 and5.0mmfromthephantomsurface(Fig.4).Hereweacquired the2-DMRimagesperpendiculartotheorientationofthepillar
Fig.4. SchematicrepresentationoftheMRIimagingsetup(notdrawntoscale).A phantomwithtwosetsofpillars,whereeachsethas13evenlydistributedpillars with200-mdiameterandsetsareseparatedby5mmapartfromeachother,is locatedinsidea12-channelheadcoil.Theproposedmetamaterialispositionedon topofthephantomtoimagethepillarsinitsvicinity.
array(i.e.,transverseplaneistheXYplanefor ˆz-directedpillars)by usingspoiledgradientecho(GRE)imagingsequencewithanEcho Time(TE)/RepetitionTime(TR)of9.25ms/100ms,afieldofview (FOV)of34×34mm,animagingmatrixof512×512,animaging durationof4min19s,apixelbandwidthof180Hz,aflipangleof 3◦andaslicethicknessof2mm.Thisimagingconfiguration cor-respondstoabout66mspatialresolutionattheimagingplane. Alloftheexperimentswereconductedbyusingstandardtwelve channelheadcoilsofa3TSiemensMagnetomTim-Trioscanner.
Theworkingprincipleoftheproposedmetamaterialstructure canbeexplainedintwo parts:1)theinductivecouplingofthe transmit-field(bodycoilsforthis imagingconfiguration)and2) thereceive-fieldinductivecoupling.Thisstructureconcentrates thefluxinitsvicinitythatistransmittedduringtheexcitationstage, whichiscalledtransmit-fieldcoupling.Inresponsetothis over excitation,excitedspinsinducessurfacecurrentsonthis metama-terialandthemetamaterialinductivelytransmitsthesesignalsto thereceivercoils(headcoilsforthisimagingconfiguration),which iscalledthereceive-fieldcoupling.
3. Resultsanddiscussions
3.1. ResonancefrequencyandQ-factorcalculations
TheEffective capacitanceof thestructureis strongly depen-dentonthedielectricthickness,tdiel,andthesurfaceareaofthe thin-filmloadedhelicaltraces.Thissurfacearea, alsocalledthe overlayregion,alsoaffectstheeffectiveinductance,Leff,because ofincreased ordecreased thenumber of turns.Partial removal of oneof the layers (e.g.,thetop layeris partially removed in Fig.1)resultsin decreased Leff and Ceff,thereby increasingthe increasedresonancefrequency.Theeffectofpartialthin-film load-ingisnumericallyanalyzedfordifferentdielectricthicknessesby usingCSTMicrowaveStudio.TheseresultsaredepictedinFig.5. Theresonancefrequencyrangeof65MHzto5.5GHzisfoundto beachievableforthesamefootprintareawiththegivendielectric thicknesses(i.e.,8mm×8mmwith2.5-mthickpolyimide).
Resonancefrequencyadjustmentofself-resonantdesigns(i.e, designsthatdonotincludelumpedelements)toapredefinedMRI frequencyisnotpracticalformostofthecases.Unlikethe previ-ouslystudiedchiralmetamaterialstructures[42]orstacked-SRR typedesigns[43],heretheproposedgeometryallowsusto
eas-Fig. 5.Resonance frequency adjustment capability of the flexible design for 8mm×8mmfootprintarea(notdrawntoscale).Forfullyloaded(doubleturn) helicalringgeometry,itispossibletoreachlowerresonancefrequencies.Partial removalofthemetallizationlayer(upperlayerforthisconfiguration)resultsin increasedresonancefrequencyviadecreasedeffectiveinductanceandcapacitance. 100%percentetchedareacorrespondstoasinglelayerSRR,wheretheresonance frequencyincreasedto5.5GHz.
Fig.6.Inputimpedanceoftheproposedmetamaterial,withafootprintareaof 8mm×8mmanddielectricthicknessof2.7m,forthegivenequivalentcircuit modelforvariousdiscretizationnumbern0.Theresultsareconvergent,andeven usingn0=10providessignificantlyaccurateresults.
ilyincreasetheresonancefrequencyafteritsmanufacturing,by usingthemetallizationremovalmethodforparallelplatecoils[18]. AsdepictedinFig.5,removalofthemetaloverlayresultsinboth decreasedthin-filmcapacitancebetweenconsecutivemetal lay-ersanddecreasedeffective-inductanceduetoloweredthenumber ofturns.Hence,theresonancefrequencyoftheproposed meta-materialstructurecanbeconvenientlyincreasedtoapredefined resonancefrequency.
AsproposedinSection2.1,theequivalentcircuitmodelforthe givenflexiblemetastructurewithafootprintareaof8mm×8mm isanalyzedtodeterminetheresonancefrequencyandtheQ-factor forvariousnumberofdiscretization.Fig.6depictstheimpedance graphsforvariousdiscretizationnumbers,n0.Theacquired reso-nancefrequenciesandQ-factorscalculatedbyusingtheFWHMof thesegraphsarealsoreportedonTable1.
Here,weobservedthattheincreasingthediscretization num-beroftheproposedarchitectureconvergestothef0of64MHzand Q-factorofapproximately46,inreasonableagreementwiththe experimentalresultsfoundas65MHzand42fortheresonance frequencyandtheQ-factorrespectively.Byusingconventional for-mulae(i.e. f0=1/(2
Lef fCef f, withL0=15.1nH, C0=303pF, andR0=0.16),wecalculatedthef0of74.4MHzandaQ-factor
Table1
Performance metricsof theproposedmetastructure,with a footprintareaof 8mm×8mmanddielectricthicknessof2.7-m,duetodifferentdiscretization number(n0). n0 f0(MHz) Q-factor Time(s) 1 45 36 0,1 2 60 44 0,1 10 64 46 0,1 100 64 46 3,1
Fig.7. Impedancegraphsforthe8mm×8mmsample.Real(top)andimaginary (bottom)partsofthecompleximpedancesshowthattheproposedequivalentcircuit modelestimatestheresonancefrequencyandtheQ-factorofthegivenstructure betterthantheconventionallumpedRLCmodelandhasaperformancesimilarto thatofthefull-wavenumericalsolutions.
of36fortheproposedmetastructure.Finally,f0of71MHzand Q-factorof44isobtainedbyusinganumericalfullwavesolver.The impedancegraphsoftheseresultsareshowninFig.7.
Similarly,forthestructurewithafootprintareaof6mm×6mm, aninductanceof9.5nHandacapacitanceof1.6nF(forthegiven silicon-nitridedielectricofrelativepermittivityof8.9andloss tan-gentof0.0006),andaresistanceof0.08isusedtocalculateits operatingfrequency.ThelumpedRLCformulationresultedinf0 of41.2MHzandQ-factorof20,whereproposedequivalentcircuit modelresultedinf0of35.4MHzandQ-factorof26,whicharein reasonableagreementwiththeexperimentalresultsof33.4MHz and13fortheresonancefrequencyandtheQ-factor,respectively. Finally,f0 of35.6MHzandQ-factorof11isobtainedbyusinga numericalfullwavesolver.Theimpedancegraphsoftheseresults arevisualizedinFig.8.
Weobservedthattheproposedequivalentcircuitmodel esti-matestheresonancefrequencyofthemetastructure,betterthan conventionalcalculationmethods. Additionally,weseethatthe proposedthin-film loadedgeometry neitherbehavesas a sim-ple series resonator, nor like a parallel resonator: rather, it is a cascaded RLC circuit with proper feedback (i.e., feedback in electricalmodelcorrespondstoaphysicalconnectioncalled cross-via-metallization)toprovide better resonancebehavior. Hence, theproposedcascadedequivalentcircuitmodelcharacterizesthe
Fig.8. Impedancegraphsforthe6mm×6mmsample.Real(top)andimaginary (bottom)partsofthecompleximpedancesshowthattheproposedequivalentcircuit modelestimatestheresonancefrequencyandtheQ-factoroftheproposedstructure betterthanconventionallumpedRLCmodelandhasaperformancesimilartothat ofthenumericalfull-wavesolver.
behaviorof thestructure moreaccuratelycomparedto conven-tional lumped-element based methods. Because the proposed circuitmodelcannotincludetheeffectofsurroundingmedium, which requiresfull-wavesimulations,it cannotestimatethe Q-factorcorrectly.Hence,calculationofQ-factorunderlossymedium, suchasloading,mightnotbereliableforthegivencircuitmodel, asisalsothecasefortheconventionallumpedRLCmodels.
In the impedance graphs presented in Fig. 8, we found the resonancefrequencyoftherigidmetamaterial(6mm×6mm res-onatorsandwichinga1-mthicksilicon-nitride)tobe33.4MHz withacorrespondingfree-spacewavelength(0)of8.98m. We calculated that thesidelength ofthe rigidresonator(6mm)is shorterthan0/1500,whichisoneofthesmallestsingle-chip deep-subwavelengthresonatorsreportedthusfarintheliterature[44]. Althoughatheoreticalworkwithasidelengthof0/1733[45]and anexperimentalworkwithlumpedcapacitorshavingasidelength of0/2000[46]werereported,thereis noexperimental demon-strationforawirelessself-resonantstructure(i.e.,withoutlumped element)electricallysmallerthanthestructurepresentedhere. 3.2. Resonancefrequencyadjustmentoftheflexiblesampletoa pre-definedmrifrequency
To achieve the resonance frequency near a predefined MRI frequency(123MHzforour3TMRIscanner)weusedoptical lithog-raphy to precisely etch the necessary amountof metal on the overlay.Weobservedthatetching17.5mm2ontheoverlay (cor-respondstoapproximately63%etchedareaofconsecutivelayers) resultedinaresonancefrequencyof126MHz,whichwasin agree-mentwiththenumericalresultsprovidedinFig.5.Themeasured Q-factoroftheflexiblemetamaterialwasalsoincreasedfrom42 toapproximately82,whichwasinagreementwithQ=2f0Leff/R viatheincreasedresonancefrequency.Itwaspreviouslyreported thattheQ-factorofconventionalstructures,(e.g.,spirals,solenoids
Table2
Experimentalcharacterizationresultsfordifferentsamples.
Type Dielectric(thickness) EtchRatio(%) f0(MHz) Q-factor
Rigid(6×6mm2) Si
3N4(1m) 0 33.4 13
Flexible(8×8mm2) Polyimide(2.7m) 0 65.0 42
Flexible(8×8mm2) Polyimide(2.7m) 60(tuned) 126.0 82
Flexible-Loaded(8×8mm2) Polyimide(2.7m) 60(tuned) 123.5 64
Fig.9. Numericalresultsshowtheelectricfieldconfinementpropertyoftheproposedmetamaterialstructurewithafootprintareaof8mm×8mm.Theelectricfield distributionclearlyindicatesthattheE-fieldisstronglyconfinedbetweenthetopandbottomlayers,leadingtohigherQ-factorevenwhenloadedinalossymedium.(a) Amplitudeoftheelectricfieldnormalizedtotheincidentfieldalongwiththedashedlinemarkedin(b)showingthattheelectricfieldis6ordersofmagnitudehigherinthe localizedregiononresonancewithrespecttotheincidentfield.
andSRRs)donotincreaselinearlybecauseofreducedinductance (andthereducedmutualinductance)fortheconsecutiveturns[33]. However,theproposedgeometryprovidesalmostalinearQ-factor improvement,viathestrongmutualcouplingamongconsecutive layers.Thus,thisstructureisveryspecialintermsofhigherQ-factor andloweredresonancefrequencyforasmallerfootprintarea.
Thetunedresonatorwasimmersedintotheabovementioned phantom,asdescribedinSection2.3,anditswirelessimpedance wasmeasuredbyusingthesamepick-upcoilandmethod[30]. TheloadedQ-factor for this configuration wasmeasuredas 64 accordingtotheFWHMofthecompleximpedance.Theresultsare summarizedinTable2forcomparison.
3.3. Electricfieldconfinementforimprovedloadingperformance Topermituseatextremely subwavelengthfrequencies, con-ventionalresonators(such asSRRs,spiralsetc.)requirelumped capacitorstotunetheirresonancefrequencies.Theelectricfield distributionofthesestructurespillsoverfromtheirplanebecause oftheselumpedcapacitors,resultinginvulnerabilitytoexternal effectssuchasloadingwithalossytissue(seedetailsarereportedin thesupplementaryfile,SectionS3).Theuseofthedouble-layer heli-calgeometryprovidesanadvantageofelectricfieldlocalizationin thelower-lossdielectricregion(comparedtolivingtissues)that allowsforhigherQ-factor.Fig.9(a)presentsthe|E|-fieldprofile ofthemapnormalizedtoincidentfield( E˙I)alongwiththedashed linemarkedinFig.9(b),whereincidentfieldistheelectricfield intensityrecordedattheexcitationportofthesimulationdomain. Weobservedthattheelectricfieldis6ordersofmagnitudehigher inthelocalizedregionunderresonanceconditionmatchedtothe MRIfrequencycomparedtotheincidentfieldandalmost3orders ofmagnitudehighercomparedtotheoff-resonancecase.Fig.9(b) indicatestheabsolutevalueoftheelectricfield distribution(|E| map)localizedinthedielectricregionbetweenthetopand bot-tommetalliclayers.Clearlytheelectricfieldisstronglyconfined betweenthemetallizationlines;thisstrongconfinementis essen-tialtoachievinghighQ-factorevenwhenloadedinalossymedium includingbiologicaltissues.
Livingtissuesexhibithigherconductivelosses,e.g.,conductivity of1S/m,whereasthedielectricshaveradicallylower conductivi-ties,e.g.,conductivityof10−16S/mforaKapton®.Theproposed metamaterialdeviceconfinestheelectricfieldinsidethedielectric material,insteadoflivingtissues,therebydrasticallydecreasingthe dissipatedresistivepoweratthegivenfrequencywhenthe metas-tructureisplacedinalossymedia.Thispropertyhelpstomaintain theQ-factorofthemetamaterialdevice,evenwhenthe metastruc-tureisloadedinvivo.Hence,theproposeddeep-subwavelength metamaterialexhibitsthesepropertiesforwirelessoperationthat arecriticaltoinvivostudies.
3.4. MRIcharacterization
The proposedstructure was locatedon a coronal plane(XZ plane),whereimmersedpillarswerealignedinthe ˆz -direction, andthesamplewaslocatedinsideastandardheadcoilofthe scan-ner,asshowninFig.10(a).Transverseimages,withtheimaging parametersgiveninSection2.3,wereacquiredtocounteachand everypillarasshowninFig.10(b).Here,weobservedthatthepillar arraycanbevisibleonlyinthevicinityoftheresonator,in agree-mentwiththeB1+ mapofthewirelessmetastructureasshown inFig.10(c).TheB1+resultsshowedthattheamplitudeoftheRF magneticfieldcanbeamplifiedsignificantlyinthevicinityofthe resonator.FromtheMR imagegiven inFig.10(b),we obtained theintensityplots alongblueand redlinesmarkedonit.From Fig.10(b),weselectedaregionwithasizeof100×100pixelsthat doesnotcontainanyMRIsource(i.e.noiseregion),andcalculated theaveragenoiseleveltobeapproximately50arbitraryunits(a.u.), wherethesignalintensityintheneighboringregionofthe res-onatorappearstobeapproximately1000a.u.Theintensitycurve inblue,dropstoapproximatelynoiselevelforthecorresponding pillarsinthevicinityofthemetamaterialresonator.Wecanclearly countthenumberofdipsas13,whichisthenumberofpillarsin thearray.Theredcurve(at5mmaway)showstheintensityprofile farawayfromtheresonator(Fig.10-d).Clearly,thepillars5mm awayfromthemetamaterialarenotvisuallyseparablefromeach other.
Fig.10.MRIcharacterizationofthetunedmetamaterial,withthefootprintareaof8mm×8mmusedtoresolvetheevenlydistributedpillarswithadiameterof200m.(a) 3TSiemensMagnetomTrioimagingsystemisusedwithaheadcoil;thesystemisloadedwithabodymimickingphantomtoimagefibersimmersedinthephantom.(b)MRI imageshowingthatpillarsareclearlyvisibleandcanbecountedinthevicinityoftheresonatoralongtheblueline(at0.1mmawayfromthemetastructure),whereasthey arenotfullyresolvablealongtheredline(5mmawayfromthemetastructure).(c)B1+mapofthewirelessmetamaterialstructureshowingsignificantB1+amplificationin itsvicinity.(d)Thered(dashed)curveindicatestheimageintensitypatternat5mmawayfromthedeviceandtheblue(solid)curveindicatestheimageintensityat0.1mm awayfromthedevice.Theblue(solid)profileclearlyresolves13ofthepillars(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothe webversionofthisarticle).
Herewealsoseethatthesignalintensityisamplified approx-imatelyoneorderofmagnitude(1000/100)inthevicinityofthe metastructure.Forthepillarsclosetotheresonatormetallization, thecontrast-to-noise-ratio(CNR)reachestoapproximately15.4 (e.g.,CNR=920−150
50 ), whereCNRdropstoapproximately6.9for the6thand7thpillarsbecauseoftheincreasedSNRinthe proxim-ityofthemeta-device,verifiedbyitsB1+map.Forthepillarsaway fromtheresonator,theCNRdropstoapproximatelyunity,which ismeaningless.Toimagesmallobjects,e.g.,pillarswith200m diameter,theSNRshouldbekepthighenoughtoeliminatenoise effects.TheproposedmetamaterialstructuremanipulatestheEM fieldstronglyinitsvicinitytoincreasetheSNR,allowingforMRIof sub-mmpillarswithanunprecedentedresolutionfor3Theadcoils thatisotherwisenotpossibleusingtheseimagingparameters.
4. Conclusion
Inconclusion,weproposed,modeledanddemonstrateda wire-less compact metamaterial architecture with a side length of approximately0/1500andfrequencyadjustmentrangeof65MHz to 5.5GHz. This proposed metastructure possess a Q-factor of approximately80for123MHz.Wedemonstratedthatthis deep-subwavelength metamaterialcan beusedas anSNR-enhancing deviceforMRI.Becauseofitsstrongelectricfieldconfinement prop-ertyinitsdielectricregion,theproposedstructureexhibitshigher Q-factorseveninalossymedium(i.e.,livingtissues).Therefore, thiswirelessmetastructureisanexcellentcandidateforusein
var-iousapplicationsincludinginvivoMRimagingplatformsandsmart implants.
Acknowledgements
HVD gratefullyacknowledgessupportfromTÜBA. Thiswork ispartiallysupportedbyTurkishNationalScientificand Techno-logical Research InstituteTÜB˙ITAK-B˙IDEB. Authorsof this work gratefullyacknowledgethesupportofUNAM-National Nanotech-nology Research Center and Instituteof Materials Science and Nanotechnology.
References
[1]V.G.Veselago,Theelectrodynamicsofsubstanceswithsimultaneously negativevaluesofand,Sov.Usp.10(1968)509–514.
[2]W.Withayachumnankul,K.aruwongrungsee,A.Tuantranont,C.Fumeaux,D. Abbott,Metamaterial-basedmicrofluidicsensorfordielectric
characterization,Sens.Actuators:Phys.189(2013)233–237.
[3]A.Daliri,A.Galehdar,S.John,C.H.Wang,W.S.T.Rowe,K.Ghorbani,Wireless strainmeasurementusingcircularmicrostrippatchantennas,Sens. Actuators:Phys.184(2012)86–92.
[4]R.Melik,E.Unal,N.K.Perkgoz,C.Puttlitz,H.V.Demir,Metamaterial-based wirelessstrainsensors,Appl.Phys.Lett.95(2009)011106.
[5]A.K.Horestani,J.Naqui,Z.Shaterian,D.Abbott,C.Fumeaux,F.Martin, Two-dimensionalalignmentanddisplacementsensorbasedonmovable broadside-coupledsplitringresonators,Sens.Actuators:Phys.210(2014) 18–24.
[6]R.Melik,E.Unal,N.K.Perkgoz,C.Puttlitz,H.V.Demir,Flexiblemetamaterials forwirelessstrainsensing,Appl.Phys.Lett.95(2009)181106.
[7]A.Alipour,E.Unal,S.Gokyar,H.V.Demir,Developmentofa distance-independentwirelesspassiveRFresonatorsensorandanew
telemetricmeasurementtechniqueforwirelessstrainmonitoring,Sens. Actuators:Phys.255(2017)87–93.
[8]M.Lapine,D.Powell,M.Gorkunov,I.Shadrivov,R.Marqués,Y.Kivshar, Structuraltunabilityinmetamaterials,APL95(2009)084105,http://dx.doi. org/10.1063/1.3211920.
[9]F.Bilotti,L.Nucci,L.Vegni,AnSRRbasedmicrowaveabsorber,Microw.Opt. Technol.Lett.48(11)(2006)2171–2175.
[10]J.B.Pendry,NegativerefractionmakesaperfectLens,Phys.Rev.Lett.85 (2000)3966.
[11]N.Fang,H.Lee,C.Sun,X.Zhang,Sub–Diffraction-limitedopticalimagingwith asilversuperlens,Science308(2005)534.
[12]M.J.Freire,R.Marques,Near-fieldimaginginthemegahertzrangebystrongly coupledmagnetoinductivesurfaces:experimentandabinitioanalysis,J. Appl.Phys.100(2006)063105.
[13]M.C.K.Wiltshire,J.B.Pendry,I.R.Young,D.J.Larkman,D.J.Gilderdale,J.V. Hajnal,MicrostructuredmagneticmaterialsforRFfluxguidesinmagnetic resonanceimaging,Science291(2001),849–51.
[14]M.J.Freire,R.Marques,L.Jelinek,Experimentaldemonstrationofau=-1 metamateriallensformagneticresonanceimaging,Appl.Phys.Lett.93(2008) 231108.
[15]E.Moser,M.Meyerspeer,F.Ph.S.Fischmeister,G.Grabner,H.Bauer,S.Trattnig, Windowsonthehumanbody-invivohigh-fieldmagneticresonanceresearch andapplicationsinmedicineandpsychology,Sensors10(2010)5724–5757. [16]M.Blaimer,F.Breuer,M.Mueller,R.M.Heidemann,M.A.Griswold,P.M.Jakob,
SMASH,SENSE,PILS,GRAPPA,howtochoosetheoptimalmethod,TopMag. Reson.Imaging15(August)(2004)223–236,Nu.4.
[17]M.A.Griswold,P.M.Jakob,R.M.Heidemann,M.Nittka,V.Jellus,J.Wang,B. Kiefer,A.Haase,Generalizedautocalibratingpartiallyparallelacquisitions (GRAPPA),Magn.Reson.Med.47(2010)1202–1210.
[18]P.Gonord,S.Kan,A.Leroy-Willig,ParallelplateSplit-conductorsurfacecoil: Analysisanddesign,Magn.Reson.Med.6(1998)353–358.
[19]P.Gonord,S.Kan,A.Leroy-Willig,C.Wary,Multigapparallel-platebracelet resonatorfrequencydeterminationandapplications,Rev.Sci.Instrum.65 (1994)3363,http://dx.doi.org/10.1063/1.1144573.
[20]T.Dohi,K.Matsumoto,I.Shimoyama,Flexiblemicroresonatorforthe magneticresonancecatheter,in:The13thInt.Conf.onSolid-StateSensors, ActuatorsandMicrosystems,Seoul,Korea,June5-9,2005.
[21]T.Matsunaga,Y.Matsuoka,S.Ichimura,Q.Wei,K.Kuroda,Z.Kato,M.Esashi, Y.Haga,Multilayeredreceivecoilproducedusinganon-planar
photofabricationprocessforanintraluminalmagneticresonanceimaging, Sens.Actuators:Phys.(2018),http://dx.doi.org/10.1016/j.sna.2017.04.021. [22]J.Gawlitza,M.Reiss-Zimmermann,G.Thörmer,A.Schaudinn,N.Linder,N. Garnov,L.C.Horn,D.H.Minh,R.Ganzer,J.U.Stolzenburg,T.Kahn,M.Moche, H.Bussea,Impactoftheuseofanendorectalcoilfor3TprostateMRIon imagequalityandcancerdetectionrate,Nat.,Sci.Rep.7(2017)40640,http:// dx.doi.org/10.1038/srep40640.
[23]C.Baltes,N.Radzwill,S.Bosshard,D.Marek,M.Rudin,MicroMRIofthe mousebrainusinganovel400MHzcryogenicquadratureRFprobe,NMR Biomed.22(2009)834–842,http://dx.doi.org/10.1002/nbm.1396.
[24]D.Sakellariou,G.LeGoff,J.F.Jacquinot,High-resolution,high-sensitivityNMR ofnanolitreanisotropicsamplesbycoilspinning,Nature447(7June)(2007) 694–697.
[25]I.Lin,H.C.Yang,J.H.Chen,Diffusiontensorimagingusingahigh-temperature superconductingresonatorina3Tmagneticresonanceimagingfora spontaneousratbraintumor,Appl.Phys.Lett.102(2013)063701. [26]E.M.Kardoulaki,R.R.A.Syms,I.R.Young,M.Rea,W.M.W.Gedroyc,Thin-film
micro-coildetectors:ApplicationinMR-thermometry,Sens.Actuators:Phys. 226(2015)48–58.
[27]R.R.R.A.Syms,I.R.Young,M.M.Ahmad,M.Rea,C.A.Wadsworth,S.D. Taylor-Robinson,Thin-filmdetectorsystemforinternalmagneticresonance imaging,Sens.Actuators:Phys.163(2010)15–24.
[28]E.Laistler,M.Poirier-Quinot,S.A.Lambert,R.M.Dubuisson,O.M.Girard,E. Moser,L.Darrasse,J.C.Ginefri,InvivoMRimagingofthehumanskinat subnanoliterResolutionusingasuperconductorsurfacecoilat1.5tesla,J. Magn.Reson.Imaging41(February(2))(2015)496–504,http://dx.doi.org/10. 1002/jmri.24549.
[29]P.A.Bottomley,TurninguptheheatonMRI,J.Am.Coll.Radiol.5(7)(July 2008)853–855.
[30]J.C.Ginefri,A.Rubin,M.Tatoulian,M.Woytasik,F.Boumezbeur,B.Djemai,M. Poirier-Quinot,F.Lethimonnier,L.Darrasse,E.Dufour-Gergam,Implanted, inductively-coupled,rfcoilsfabricatedonflexiblepolymericmaterial: ApplicationtoinvivoratbrainMRIat7T,J.Magn.Reson.224(2012)61–70. [31]J.A.Lehmann-Horn,J.F.Jacquinot,J.C.Ginefri,C.Bonhomme,D.Sakellariou,
MonolithicMACSmicroresonators,J.Magn.Reson.271(2016)46–51. [32]R.Kriegl,J.C.Ginefri,M.Poirier-Quinot,L.Darrasse,S.Goluch,A.Kuehne,E.
Moser,E.Laisler,Novelinductivedecouplingtechniqueforflexible transceiverarraysofmonolithictransmissionlineresonators,Magn.Reson. Med.73(2015)1669–1681.
[33]F.Bilotti,A.Toscano,L.Vegni,DesignofspiralandmultipleSplit-ring resonatorsfortherealizationofminiaturizedmetamaterialsamples,IEEE Trans.AntennasPropag.55(August(8))(2007),http://dx.doi.org/10.1109/ TAP.2007.901950.
[34]D.R.Smith,J.B.Pendry,M.C.K.Wiltshire,MetamaterialsandNegative RefractiveIndex,Science305(2004)788.
[35]Q.Y.Tang,Y.M.Pan,Y.C.Chan,K.W.Leung,Frequency-tunablesoftcomposite antennasforwirelessstrainsensing,Sens.Actuators:Phys.179(2012) 137–145.
[36]J.D.Baena,J.Bonache,F.Martin,R.MarquesSillero,F.Falcone,T.Lopetegi, M.A.G.Laso,J.Garcia-Garcia,I.Gil,M.F.Portillo,M.Sorolla,Equivalent-circuit modelsforsplit-ringresonatorsandcomplementarysplit-ringresonators coupledtoplanartransmissionlines,IEEETrans.Microw.TheoryTechnol. MTT-53(2005)1451–1461.
[37]S.S.Mohan,M.Hershenson,Simpleaccurateexpressionforplanarspiral inductances,IEEEJ.SolidStateCircuits34-10(October)(1999)1419–1424. [38]InderBahl,LumpedElementsforRFandMicrowaveCircuits”,ArtechHouse,
2003.
[39]J.C.Ginefri,E.Durand,L.Darrasse,Quickmeasurementofnuclearmagnetic resonancecoilsensitivitywithasingle-loopprobe,Rev.Sci.Instrum.70 (1999)4730.
[40]A.L.Coutrot,E.Dufour-Gergam,E.Martincic,J.P.Gilles,J.P.Grandchamp,J.M. Quemper,A.Bosseboeuf,F.Alves,B.Ahamada,Electromagneticmicro-device realizedbyelectrochemicalway,Sens.ActuatorsA91(2001)80–84. [41]V.Acikel,O.Ulutan,A.C.Ozen,B.Akin,Y.Eryaman,E.Atalar,AnovelMRI
basedelectricalpropertiesmeasurementtechnique,Proc.Intl.Soc.Mag. Reson.Med.(2013)21.
[42]R.Marques,L.Jelinek,F.Mesa,Negativerefractionfrombalanced quasi-planarinclusions,Microw.Opt.Technol.Lett.49(10)(2007). [43]E.Ekmekci,K.Topalli,T.Akin,G.Turhan-Sayan,Atunablemulti-band
metamaterialdesignusingmicro-splitSRRstructures,Opt.Express17(18) (2009).
[44]FabriceLemoult,NadegeKaina,MathiasFink,GeoffroyLerosey,Wave propagationcontrolatthedeepsubwavelengthscaleinmetamaterials,Nat. Phys.9(2013).
[45]X.Zhang,E.Usi,S.K.Khan,M.Sadatgol,D.O.Guney,Extremely
sub-wavelengthnegativeindexmetamaterial,Prog.Electromagnet.Res.152 (2015)95–104.
[46]C.P.Scarborough,Z.H.Jiang,D.H.Werner,C.Rivero-Baleine,C.Drake, Experimentaldemonstrationofanisotropicmetamaterialsuperlenswitha negativeunitypermeabilityat8.5MHz,Appl.Phys.Lett.101(2012)014101.
Biographies
Sayim GokyarcompletedhisPh.D. atBilkent Univer-sity,DepartmentofElectricalandElectronicsEngineering, Ankara.Hisresearchinterestsincludewirelesssensors anddesigningimplantableelectronicdevicesforwireless imagingapplicationsaswellasforsensing.
AkbarAlipourcompletedhisPh.D.atBilkent Univer-sity,DepartmentofElectricalandElectronicsEngineering, Ankara.Hisresearchinterestslieintheareaofthin-film microwavestructureswhichareusedformagnetic res-onanceimaging(MRI)markingandwirelesssensing.He hasbeeninvolvedinthedevelopmentofdeviceswhichare mainlyusedininterventionalMRIandmedicalimplants.
EmreUnalreceivedhisB.S.degreeinelectricaland elec-tronicsengineeringfromHacettepeUniversity,Ankara, Turkey,in2005.Heisafull-timeResearchEngineerunder thesupervisionofProf.H.V.DemirwiththeInstituteof MaterialsScienceandNanotechnology,Bilkent Univer-sity,Ankara,whereheisworkingonthedevelopmentof microwaveandoptoelectronicdevices.
ErginAtalarreceivedhisB.S.degreefromBogazici Uni-versityin1985,M.S.degreefromMiddleEastTechnical Universityin1987,andPh.D.degreefromBilkent Univer-sityin1991,allinElectricalEngineering.Hejoinedthe JohnsHopkinsUniversity,wherehebecameaProfessor ofRadiology,BiomedicalEngineeringandElectricaland ComputerEngineeringandDirectorofCenterforImage GuidedInterventions.Currently,Dr.AtalarisaProfessorof theDepartmentofElectricalandElectronicsEngineering andtheDirectorofNationalMagneticResonanceResearch Center atBilkent University.Themainresearch inter-estsofDr.AtalarareMagneticResonanceImagingand Imageguidedinterventions.In2006,ErginAtalarwonthe TUBITAKscienceaward.
Hilmi Volkan Demir (M’04-SM’11) received the B.S. degree inelectrical and electronics engineeringfrom BilkentUniversity,Ankara,Turkey,in1998andtheM.Sc. andPh.D.degreesinelectricalengineeringfromStanford University,Stanford,CA,in2000and2004,respectively. InSeptember2004,hejoinedBilkentUniversity,where heiscurrentlyaprofessorwithjointappointmentsatthe DepartmentofElectricalandElectronicsEngineeringand theDepartmentofPhysicsandisalsowithUNAM-the InstituteofMaterialsScienceandNanotechnology. Con-currently,heisafellowofNationalResearchFoundation inSingaporeandaprofessorofNanyangTechnological University.Hisresearchinterestsincludethe develop-mentofinnovativeoptoelectronicandRFdevices.Prof.Demiristherecipientof theEuropeanUnionMarieCurieFellowship,theTurkishNationalAcademyof Sci-encesDistinguishedYoungScientistAward(TUBA-GEBIP),theEuropeanScience Foundation-EuropeanYoungInvestigatorAward(ESF-EURYI),andNanyangAward forResearchExcellence.